1. Field of the Invention
The present invention relates to a semiconductor memory device, and more particularly to a semiconductor memory device with a hierarchical bit line architecture having main and sub bit lines.
2. Description of the Related Art
Mask Read Only Memory (MROM) is a conventional semiconductor memory device which adopts a hierarchical bit line architecture. MROM includes a memory cell structure having two transistors (MOSFET) with different level threshold levels so as to store a binary digit. One of the two transistors has a low threshold level, i.e., a low voltage applied to the gate electrode can cause a sufficient current to flow through the transistor. The other has a high threshold level such that only a sufficiently high voltage applied to the gate electrode allows the current flow. The use of two transistors having low and high threshold levels, respectively, can realize the binary memory. The hierarchical bit line architecture is used to read the value of the binary digit stored in the corresponding transistor, and to integrate the memory cells in high density.
FIG. 2 is a circuit diagram representing an MROM with a hierarchical bit line architecture. As shown in FIG. 2, main bit lines MB1, MB2, . . . , etc. and sub bit lines SB1, SB2, . . . , etc. are arranged in a hierarchical structure. For example, the main bit line MB1 is disposed between the odd-numbered sub bit line SB1 and the even-numbered sub bit line SB2. One of memory cells M1, M2, . . . , etc., which includes a transistor, is disposed each between the adjacent two sub bit lines. The source of each of the memory cells M1, M2, . . . , etc. is connected to one of the adjacent two sub bit lines and the drain is connected to the other. The gates of the memory cells are connected to word lines WL0, . . . , WLn. The main bit lines MB1, MB2, . . . etc. are connected to sense circuits SA1, SA2, . . . , etc., respectively, and a charging circuit Ca, or are otherwise connected via transistors (MOSFET) Q1, Q2, . . . , etc. to GND. Each odd-numbered sub bit line is connected via a bank select transistor (MOSFET) BK1 or BK2 to a main bit line. The gates of the bank select transistors (MOSFET) BK1 and BK2 are connected to bank select lines BKL1 and BKL2, respectively. Each even-numbered sub bit line is connected via a bank select transistor (MOSFET) BK3 or BK4 to a main bit line. The gates of the bank select transistors BK3 and BK4 are connected to bank select lines BKL3 and BKL4, respectively.
In order to read the value of the memory cell M4, for example, the bank select line BKL1 is activated so that the bank select transistor BK1 is switched ON while the bank select line BKL3 is activated so that the bank select transistor BK3 is switched ON. This leads to formation of a current path from the charging circuit Ca to the main bit line MB1 to the bank select transistor BK1 to the sub bit line SB3. When the threshold of the memory cell M4 is set to a low value, the activated word line WL0 switches ON the memory cell M4. This allows a current to flow from the sub bit line SB3, to the memory cell M4, to the sub bit line SB4, to the bank select transistor BK3, to the main bit line MB2, to the transistor Q1, to GND. On the other hand, when the threshold of the memory cell M4 is set to a high value, the activated word line WL0 leaves the memory cell M4 OFF. This does not allow a current to flow in the above-described path. The sense circuit SA1 reads the binary digit stored in the memory cell M4 by determining whether a current flows through the path or not.
FIG. 3 is a diagram illustrating a layout design of the semiconductor substrate on which the MROM shown in FIG. 2 is provided.
In FIG. 3, a channel region Ch is provided in each of the bank select transistors BK1, BK2, . . . , etc. An auxiliary conduction region (diffusion region) H is connected via each of the bank select transistors BK1, BK2, . . . , etc. to each of the sub bit lines SB1, SB2, . . . , etc. Each of the bank select transistors BK1, BK2, . . . , etc. is connected via a contact hole CC of the corresponding auxiliary conduction region H to each of the main bit lines MB1, MB2, . . . , etc. The word lines WL0, . . . , WLn intersect the sub bit lines SB1, SB2, . . . , etc.
The main bit lines MB1, MB2, . . . , etc. are made of a low resistance material such as metal. The sub bit lines SB1, SB2, . . . , etc. are formed of a diffusion layer which enables formation of the source and drain of a transistor. Each memory cell uses two adjacent sub bit lines as a source electrode and a drain electrode, respectively, and one of the word lines WL0, . . . , WLn as a gate electrode in a way shown in FIG. 3.
Here the bank select transistors BK1, BK2, etc. each have channel region Ch having a large width w2. The large-width channel regions Ch increase current flowing from the charging circuit to the memory cells M1, M2, . . . , etc. to GND, and the increased current can improve the performance of the bank select transistors BK1, BK2, . . . , etc. This leads to high-speed reading of data from the memory cells M1, M2, . . . , etc.
As shown in FIG. 3, the channel regions Ch of two adjacent bank select line s BK1, BK2, . . . , etc. face each other. For example, the channel region Ch of the bank select line BK1 faces the channel region Ch of the bank select line BK2. The channel region Ch of the bank select line BK3 faces the channel region Ch of the bank select line BK4. The width b between the bank select transistors, e.g. , between BK1 and BK2 or between BK3 and BK4, is required to sufficiently isolate the bank selects transistors. The sufficient width b leads to an increase in the width a of a memory array, resulting in a large-sized chip.
To solve such a problem, Japanese Laid-Open Publication No. 6-104406 discloses an MROM which has a circuit structure and layout design as shown in FIGS. 4 and 5, respectively. Like numeral references refer to like parts in FIGS. 2 through 5.
As is apparent from FIG. 4, bank select transistors BK2-1 and BK2-2 are connected to each other in parallel and bank select transistors BK3-1 and BK3-2 are connected to each other in parallel. Each of the bank select transistors BK2-1 and BK2-2 has the same current supply performance (current amount per unit area) as that of the bank select transistor BK2 shown in FIG. 2. Each of the bank select transistors BK3-1 and BK3-2 has the same current supply performance as that of the bank select transistor BK3 shown in FIG. 2.
As is apparent from FIG. 5, the auxiliary conduction regions H (diffusion region) each have an H-shape. A sub bit line SB1 is connected to an auxiliary conduction region H via the bank select transistors BK2-1 and BK2-2. Similarly, sub bit line SB2 is connected to an auxiliary conduction region H via the bank select transistor BK4. Sub bit line SB3 is connected to an auxiliary conduction region H via the bank select transistor BK1. Sub bit line SB4 is connected to an auxiliary conduction region H via the bank select transistors BK3-1 and BK3-2. The auxiliary conduction regions H each are connected via a contact hole CC of an insulating layer (not shown) to one of the main bit lines MB1, MB2, . . . , etc.
Here the width w1 of the channel region Ch of each of the bank select transistors BK2-1 and BK2-2 and the bank select transistors BK3-1 and BK3-2 is smaller than the width w2 of the channel region Ch of each of the bank select transistors BK2 and BK3 shown in FIG. 3. This prevents an increase in chip size.
The width W1 of the channel region Ch of each of the bank select transistors BK2-1 and BK2-2 may be small. However, when the sum of the widths w1 of the channel regions Ch of the bank select transistors BK2-1 and BK2-2 (i.e., w1+w1) is equal to the width w2 of the channel region Ch of the bank select transistor BK2 shown in FIG. 3 (w1+w1=w2), the bank select transistors BK2-1 and BK2-2 provide the same amount of current flow as that of the bank select transistor BK2 shown in FIG. 3. The same applies to the relationship between the bank select transistors BK3-1 and BK3-2 and the bank select transistor BK3 shown in FIG. 3.
Even when the relationship w1+w1=w2 is satisfied in the layout design shown in FIG. 5, the relationship does not hold in the actual LSI due to the microstructure. In this case, the bank select transistors BK2-1 and BK2-2 shown in FIG. 4 do not have the same current supply performance as that of the bank select transistor BK2 shown in FIG. 3. The bank select transistors BK3-1 and BK3-2 shown in FIG. 4 do not have the same current supply performance as that of the bank select transistor BK3 shown in FIG. 2. This is because the effective width of the channel region Ch of a bank select transistor is smaller in the actual LSI than in the layout design. In other words, the width of the channel region Ch of the bank select transistor in the layout design includes an ineffective width which does not function as the channel region Ch. This ineffective width does not depend on the gate width when the gate length is constant.
FIG. 6 shows a channel region Ch of a bank select transistor between two sub bit lines SB. FIG. 6 also shows the relationship between the width w1 of the channel region Ch in a layout design and the effective and ineffective widths w6 and w4 of the channel region Ch in the actual LSI (w6=w1xe2x88x922xc3x97w4).
Taking the relationship into consideration, the sum Weff0 of the effective widths of the channel region ch of the bank select transistors BK2-1 and BK2-2 is represented by the following equation: Weff0=(w1xe2x88x922xc3x97w4)+(w1xe2x88x922xc3x97w4)=2xc3x97w1xe2x88x924xc3x97w4. The effective width Weff0 of the channel region Ch of the bank select transistor BK2 shown in FIG. 3 is represented by the following equation: Weff1=w2xe2x88x922xc3x97w4=2xc3x97w1xe2x88x922xc3x97w4 where w2 is the width of the channel region Ch of the bank select transistor BK2 in the layout design (=2xc3x97w1). Since the ineffective width w4 is constant as described above, the sum Weff0 of the effective widths of the channel region oh of the bank select transistors BK2-1 and BK2-2 shown in FIG. 5 is smaller than the effective width Weff1 of the channel region Ch of the bank select transistor BK2 shown in FIG. 3. Moreover, the value (w1+w1=w2) defined in the layout design often does not hold in the actual LSI due to variations in the width w1. For these reasons, the combinations of the bank transistors BK2-1 and BK2-2 as well as BK3-1 and BK3-2 have reduced current supply performance.
The width w1 of the channel region Ch in the layout design can be defined to allow for the ineffective width w4. The ineffective width w4 varies depending on the manufacturing conditions of the actual LSI, so that it is difficult to control the effective width w6. Variations in the bank select transistors cause the reading speed of the memory cell to vary among the bank select transistors through which the value in a memory cell is read out. For this reason, a stable and fast reading cannot be established.
Furthermore, the auxiliary conduction region H has a parasitic capacitance. As is apparent from the layout design shown in FIG. 5, the auxiliary conduction region H is H-shaped. The large area of the auxiliary conduction region H leads to an increase in the parasitic capacitance of the main bit line. This results in the slow speed at which the memory cell value is read out.
According to one aspect of the present invention, a semiconductor memory device includes a semiconductor substrate; a plurality of word lines provided on the semiconductor substrate and arranged in parallel to each other; a plurality of memory cells provided along each of the plurality of word lines; a plurality of sub bit lines provided on the semiconductor substrate and arranged in parallel to each other, each of the plurality of word line intersecting the plurality of sub bit lines; a plurality of main bit lines arranged in parallel to the plurality of sub bit lines; a plurality of bank select lines arranged in parallel to the plurality of word lines; a plurality of bank select transistors provided along each of the plurality of bank select lines and connected to the respective sub bit lines; and a plurality of auxiliary conduction regions provided for each of the plurality of the main bit lines, connecting each of the plurality of the main bit lines to a set of the plurality of bank select transistors of the plurality of sub bit lines. The shape of the plurality of bank select transistors is the same. Each of the plurality of auxiliary conduction regions has a modified-H shape including a central portion connected to each of the plurality of main bit lines and branch portions each connected to a corresponding one of the set of the plurality of bank select transistors of the plurality of sub bit lines.
In one embodiment of this invention, each of the plurality of auxiliary conduction regions has four branch portions and has a modified-H shape.
In one embodiment of this invention, two of the four branch portions are smaller than the others.
In one embodiment of this invention, the semiconductor substrate is of a first conductivity type and the plurality of sub bit lines are a second conductivity type diffusion layer.
In one embodiment of this invention, each of the plurality of memory cells has one of the plurality of sub bit lines as a source and a drain and one of the plurality of word lines as a gate electrode.
In one embodiment of this invention, each of the plurality of main bit lines is made of a low resistance metal material.
According to another aspect of the present invention, a semiconductor memory device includes a semiconductor substrate; a plurality of word lines provided on the semiconductor substrate and arranged in parallel to each other; a plurality of memory cells provided along each of the plurality of word lines; a plurality of sub bit lines provided on the semiconductor substrate and arranged in parallel to each other, each of the plurality of word line intersecting the plurality of sub bit lines; a plurality of main bit lines arranged in parallel to the plurality of sub bit lines; a plurality of bank select lines arranged in parallel to the plurality of word lines; a plurality of bank select transistors provided along each of the plurality of bank select lines and connected to the respective sub bit lines; and a plurality of auxiliary conduction regions provided for each of the plurality of the main bit lines, connecting each of the plurality of the main bit lines to a set of the plurality of sub bit lines. The plurality of bank select transistors provided along one of the plurality of bank select lines and the plurality of bank select transistors provided along another of the plurality of bank select lines adjacent to the one of the plurality of bank select lines are alternately arranged and protrude inwardly between the one and another of the plurality of word lines.
In one embodiment of this invention, the semiconductor substrate is of a first conductivity type and the plurality of sub bit lines are a second conductivity type diffusion layer.
In one embodiment of this invention, each of the plurality of memory cells has one of the plurality of sub bit lines as a source and a drain and one of the plurality of word lines as a gate electrode.
In one embodiment of this invention, each of the plurality of main bit lines is made of a low resistance metal material.
Thus, the invention described herein makes possible the advantage of providing a semiconductor memory device providing stable high-speed reading without an increase in chip size.